Here we describe methods for preclinical evaluation of candidate medications to treat opioid abuse and dependence. Our perspective is founded on the propositions that (1) drug self-administration procedures provide the most direct method for assessment of medication effects, (2) procedures that assess choice between opioid and nondrug reinforcers are especially useful, and (3) the states of opioid dependence and withdrawal profoundly influence both opioid reinforcement and the effects of candidate medications. Effects of opioid medications on opioid choice in nondependent and opioid-dependent subjects are reviewed. Various nonopioid medications have also been examined, but none yet have been identified that safely and reliably reduce opioid choice. Future research will focus on (1) strategies for increasing safety and/or effectiveness of opioid medications, and (2) continued development of nonopioids such as inhibitors of endocannabinoid catabolic enzymes or inhibitors of opioid-induced glial activation.